Centralized and seamless, the cloud is touted by providers and consultants as the key to transforming business models with advanced technologies and enterprise IT.
However, chip makers are demonstrating a similar capacity for disruptive development as they reach outward from the data center to better utilize computing power in the end-point devices populating the edge of the connected systems that comprise the Internet of Things.
Leveraging the architectures and deepening densities of the data their chips can process is helping chip makers meet demand from businesses and consumers for all kinds of devices, capable of voice and object recognition and for predictive applications, like those at work in self-driving vehicles.
Coupled with the impending launch of 5G communications technology that will greatly expand the number of connected devices when rollout begins in 2020, the push to deliver chips to market that deliver computing power at the edge is gaining pace.
Stealing a March
That’s prompting some chip makers to try to steal a march both on competition within the sector and from companies such as Amazon Web Services, Oracle and Microsoft that dominate the market for cloud services.
Intel’s release late last month of a toolkit that extracts more from the 1.6 exabytes of data that can be captured daily by a single video camera is indicative of the focus on the network periphery.
Designed for use in environments ranging from manufacturing to retail, the giant chip maker’s Open Visual Inference and Neural Network Optimization – OpenVINO – helps onboard processors parse image data from the edge and send only actionable insights to the cloud.
Accommodating Machine Learning
Applying artificial intelligence to sift through large datasets from a diverse array of IoT sources is a hallmark of the advanced technologies enabled by the cloud.
To facilitate them, chip architectures also must accommodate the machine learning that runs in parallel with analytics engines at the heart of AI in order to augment algorithms based on the data flows they handle.
Product families introduced in February by chip maker AMD bridge cloud and edge by running multiple processes concurrently with data in differing formats. Designed to be embedded in servers and devices, these capabilities improve connectivity and enable the handling of higher volumes of data at faster speeds.
Networking giant Cisco Systems is using the company’s EPYC chips to reduce energy costs for servers, in part through the distribution of computing power through what it calls “micro data centers†that can be controlled from the cloud. Meanwhile, AMD’s Ryzen chips marry core and graphics architectures for edge deployments in imaging and gaming, taking aim at market leader Intel in PCs.
Connected Devices
Also seeking to dominate at the edge is Qualcomm, the US maker whose chips power smartphones and radio communications devices. Their number is set to grow significantly past the estimated 21 billion connected devices currently in use worldwide when rollout of 5G and its wider frequency spectrum gets underway in the US in about 18 months.
Unveiled at last week’s Augmented World Expo event in Silicon Valley, the company’s XR1 chip unites virtual and augmented realty technologies to deliver ultra high-definition 4K video to smartphones and gaming consoles.
Days before, the company announced it would provide chips and software integration for Facebook’s Terragraph initiative, which aims to deliver fast WiFi without the need for fiber optic cabling as early as next year.
Chipset as a Service
Those developments are in line with Qualcomm’s vision for 5G, which it seeks to realize through the implementation of higher-capacity data processing capabilities for mobile devices.
In February, the company – which has refined its business line over the past decade from the production of cellular phones to focus on chips – announced it was implementing a chipset-as-a-service model that combines lifecycle management and network security.
The Wireless Edge Services offering is aimed at industrial IoT users of 4G and 5G standards, allowing them to roll out devices at scale and benefit from outsourced management of onboarding, activation, maintenance and upgrading.
Initial applications in industrial, automotive and home security environments are set to debut later this year, with Qualcomm’s subscription revenue a testament to the disruptive potential at the edge of the cloud’s transformative power.